Simple True White Noise Generator

[Sigurthr]

As referenced to in the TCBT (tesla coil build thread). I decided to post it here instead of buried in the TCBT because it was O/T there and it should be more accessible for others.

http://img27.imageshack.us/img27/7951/og2c.png

Simple, effective, and free of the annoying pops/repeats that pseudo-white noise sources have.

 

[jcranmer]

Looks interesting and easy to build. Might be time for me to sample some components again. (Or source some cheap)

 

[Sigurthr]

I forgot to include on it that R4 is the Gate bias adjustment to put Q3 into linear operation. You don't even need a scope as you can just adjust R4 for maximum output volume. I can suggest a good Q3 if needed but any N-channel enhancement mode MOSFET should work. The gain achieved will vary with MOSFET used, of course, but since it is being followed up with an LM-386 it shouldn't be a big deal. Volume adjustment can be done by severing the connection between C4 and R1 and adding a 10K linear potentiometer: wiper to R1, Right pin (shaft towards you) to the negative side of C4, Left pin to to Pin 2 (GND) of the LM386.

 

[The Lightning Stalker]

Wow, that is much simpler than what I ended up building. It's a shame this wasn't available
at the time. It works on a single ended supply as well. Mine needs +9 0 -9 to operate, the
main reason why it's still on the breadboard. I still don't have a separate power supply to run
it with.

Would a 2N7000 work?

 

[Sigurthr]

I haven't tried it with other NPN transistors, but the EE books say that it "should". I'm not much of a bipolar transistor guy. Besides, 2N2222's are cheap as hell and abundant the world over.

All of my designs favor simplicity and low component counts, I'm too poor to do anything else, haha.

Also, this circuit will directly power a passive speaker (no need for the powered computer speakers like in your pic). Depending on the version of LM386 used you can get upwards of 1W of power too.

 

[The Lightning Stalker]

The 2N7000 is a TO-92 enhancement mode MOSFET. I figured it would be the first thing you would try.

There is nothing wrong with your design. It's a great design. I just like having the flexibility and modularity of an audio jack output. That way I can use it for other things like testing audio equipment. (Pink noise instead of white) When it's more built out, it will also have RCA outputs and possibly others, if that ever happens.

 

[Sigurthr]

Ah, my mistake. Never even heard of the 2N7000 before you mentioned it, heh. I believe it has to be a bipolar transistor pair to generate white noise, as it is driven by electron leakage across the reverse biased diode junction.

FYI; my "go-to" fet is the FDL100N50F or the FDP8874 for lower loads. I hardly ever have to get/use anything else.

 

[The Lightning Stalker]

I was referring to Q3, yeah. It would probably make a good candidate for Q3. Nice and small, compact for
small signal amplification

The filter stage to convert from white to pink would go after the MOSFET, I suppose.

 

[Sigurthr]

Oh, gotcha! Heh, sorry, I'm a bit frazzled after hours of tripple checking my SSTC driver and pulling up component info. Ya that should work fine for Q3, it's just a basic class-A voltage gain amp.

The 386 has 2MHz of bandwidth so you could even filter after the output depending on needs.

 

[Flaminpyro]

What are you guys using the white noise gen for :thinking:

going to sleep at night ???

 

[The Lightning Stalker]

Yes, and testing audio equipment, adjusting GEQs, etc.

 

[ped]

Didnt GEQ's use pink noise?

 

[Sigurthr]

What are you guys using the white noise gen for :thinking:

going to sleep at night ???

I use it for sleeping to, haha.

I still adjust GEQs with my own two ears using music with good spectral bandwidth. Not all speaker systems are created equally and IMO you need to map out their resonances and attenuations and then correct for them. The masking nature of white/pink noise would make it hard to catch these peaks and nulls.

Also, you can either send the output from the 386 to a RF amp for 0-2MHz bandwidth, or bypass/remove the 386 and send the output to a RF preamp and amp for a wideband RF signal jammer. I've measured the output from the MOSFET amp stage at up to 50MHz (limit of my scope).

 

[Atomicrox]

Interesting, might try to build one when I get some free time.

 

[phenol]

would be nice to see the spectrum of that noise source on a thing like this...

we got this analyzer at work for precompliance emi testing and it's showing some broadband noise emanating from a digital circuit board.
the other pic shows a set of near field H and E probes...

when i have time i will scan the near field emission of a typical laser switching mode driver in action...

 

[The Lightning Stalker]

I'd say build one and find out. That is a nice specan. I was looking at the Motorola R2001D the other
day which is much older and cheaper!